You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I used a VGG16 based UNet to learn to segment buildings from an image. During training, I found that the encoder's activation starts to shrink to small area of the buildings, and also show up on some background area, does this show that my model is overfitting?
Examples from epoch 5 and epoch 24:
Epoch5:
Epoch 24:
Epoch5:
Epoch24:
Epoch5:
Epoch24:
The text was updated successfully, but these errors were encountered:
I used a VGG16 based UNet to learn to segment buildings from an image. During training, I found that the encoder's activation starts to shrink to small area of the buildings, and also show up on some background area, does this show that my model is overfitting?
Examples from epoch 5 and epoch 24:
Epoch5:
Epoch 24:
Epoch5:
Epoch24:
Epoch5:
Epoch24:
The text was updated successfully, but these errors were encountered: